Table 1

Potential Predictor Variables Evaluated in 28 Therapy-Specific Logistic Regression Models

Dependent Variables for Logistic Regressions
Potential Predictor Variable
 High Knowledge of Therapy*
 Prior Use of Therapy*
 Prior Use of Therapy for Back Pain*
 High Expectations of Success of Therapy*
 Likelihood of Trying Therapy at No Cost*
 Likelihood of Trying Therapy for $10 Co-pay**
Geographic location (Boston vs. Seattle)
 X
 X
 X
 X
 X
 X
Age (65+ vs. < 65)
 X
 X
 X
 X
 X
 X
Gender (female vs. male)
 X
 X
 X
 X
 X
 X
Race (white, non-white)
 X
 X
 X
 X
 X
 X
Education (no college vs. some college)
 X
 X
 X
 X
 X
 X
≥ 5 years since first back pain
 X


X
 X
≥ 90 days of LBP in last 6 mo.
 X


X
 X
High symptom bothersomeness (7 – 10) on a 0 – 10 scale
 X


X
 X
High knowledge of therapy (4 or 5) on a 1 – 5 scale


X
 X
 X
Prior use of therapy


X
 X
 X
Prior use of therapy for back pain


X
 X
 X
High expectations of therapy (7 – 10) on a 0 – 10 scale



X
 X
Medication usage in past week



X
 X
Prior harm from therapy



X
 X

* Separate models were done for each of the five therapies (acupuncture, chiropractic, massage, meditation, t'ai chi) ** Separate models were done for acupuncture, chiropractic, and massage. An X indicates that a particular potential predictor variable was evaluated in a model with the specific dependent variable.

Sherman et al. BMC Complementary and Alternative Medicine 2004 4:9   doi:10.1186/1472-6882-4-9